1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
|
// RUN: mlir-spirv-cpu-runner %s -e main --entry-point-result=void --shared-libs=%mlir_runner_utils,%mlir_test_spirv_cpu_runner_c_wrappers \
// RUN: | FileCheck %s
// CHECK: data =
// CHECK-RAW: [[[7.7, 0, 0], [7.7, 0, 0], [7.7, 0, 0]], [[0, 7.7, 0], [0, 7.7, 0], [0, 7.7, 0]], [[0, 0, 7.7], [0, 0, 7.7], [0, 0, 7.7]]]
module attributes {
gpu.container_module,
spirv.target_env = #spirv.target_env<
#spirv.vce<v1.0, [Shader], [SPV_KHR_storage_buffer_storage_class, SPV_KHR_8bit_storage]>,
#spirv.resource_limits<
max_compute_workgroup_invocations = 128,
max_compute_workgroup_size = [128, 128, 64]>>
} {
gpu.module @kernels {
gpu.func @sum(%arg0 : memref<3xf32>, %arg1 : memref<3x3xf32>, %arg2 : memref<3x3x3xf32>)
kernel attributes { spirv.entry_point_abi = #spirv.entry_point_abi<workgroup_size = [1, 1, 1]>} {
%i0 = arith.constant 0 : index
%i1 = arith.constant 1 : index
%i2 = arith.constant 2 : index
%x = memref.load %arg0[%i0] : memref<3xf32>
%y = memref.load %arg1[%i0, %i0] : memref<3x3xf32>
%sum = arith.addf %x, %y : f32
memref.store %sum, %arg2[%i0, %i0, %i0] : memref<3x3x3xf32>
memref.store %sum, %arg2[%i0, %i1, %i0] : memref<3x3x3xf32>
memref.store %sum, %arg2[%i0, %i2, %i0] : memref<3x3x3xf32>
memref.store %sum, %arg2[%i1, %i0, %i1] : memref<3x3x3xf32>
memref.store %sum, %arg2[%i1, %i1, %i1] : memref<3x3x3xf32>
memref.store %sum, %arg2[%i1, %i2, %i1] : memref<3x3x3xf32>
memref.store %sum, %arg2[%i2, %i0, %i2] : memref<3x3x3xf32>
memref.store %sum, %arg2[%i2, %i1, %i2] : memref<3x3x3xf32>
memref.store %sum, %arg2[%i2, %i2, %i2] : memref<3x3x3xf32>
gpu.return
}
}
func.func @main() {
%input1 = memref.alloc() : memref<3xf32>
%input2 = memref.alloc() : memref<3x3xf32>
%output = memref.alloc() : memref<3x3x3xf32>
%0 = arith.constant 0.0 : f32
%3 = arith.constant 3.4 : f32
%4 = arith.constant 4.3 : f32
%input1_casted = memref.cast %input1 : memref<3xf32> to memref<?xf32>
%input2_casted = memref.cast %input2 : memref<3x3xf32> to memref<?x?xf32>
%output_casted = memref.cast %output : memref<3x3x3xf32> to memref<?x?x?xf32>
call @fillF32Buffer1D(%input1_casted, %3) : (memref<?xf32>, f32) -> ()
call @fillF32Buffer2D(%input2_casted, %4) : (memref<?x?xf32>, f32) -> ()
call @fillF32Buffer3D(%output_casted, %0) : (memref<?x?x?xf32>, f32) -> ()
%one = arith.constant 1 : index
gpu.launch_func @kernels::@sum
blocks in (%one, %one, %one) threads in (%one, %one, %one)
args(%input1 : memref<3xf32>, %input2 : memref<3x3xf32>, %output : memref<3x3x3xf32>)
%result = memref.cast %output : memref<3x3x3xf32> to memref<*xf32>
call @printMemrefF32(%result) : (memref<*xf32>) -> ()
return
}
func.func private @fillF32Buffer1D(%arg0 : memref<?xf32>, %arg1 : f32)
func.func private @fillF32Buffer2D(%arg0 : memref<?x?xf32>, %arg1 : f32)
func.func private @fillF32Buffer3D(%arg0 : memref<?x?x?xf32>, %arg1 : f32)
func.func private @printMemrefF32(%arg0 : memref<*xf32>)
}
|
